American actor Tom Hanks and television host Gayle King have warned their fans about advertisements that use their appearance and voice, without their knowledge, thanks to artificial intelligence technologies, to promote products . "Attention!! There's a video going around promoting dental insurance with an AI version of me. I had nothing to do with it,” Tom Hanks said on his Instagram profile on Saturday, October 7. His message appears on a screenshot of the video in question.
Gayle King, who co-hosts the American channel CBS's morning show, posted a similar warning on Monday, also on the Instagram platform. “People keep sending me this video and asking me questions about this product, even though I have NOTHING to do with this company...” she wrote in the caption. a video where she appears to be promoting a weight loss product. The journalist also published the original video used by the fraudsters, in which she promoted a radio show on August 31. “They manipulated my voice and the video to make it seem like I was promoting this product...I have never heard of this product and I have never used it!” she insisted, before concluding: “Don’t let yourself be fooled by these AI videos…”
The video pinned by the host appears to have been published by a certain “Artipet”, but searches on Instagram and Google yield no results referencing such diet products. For years, artificial intelligence has made it possible to produce “deepfakes”, that is to say, to manipulate images and sound, to make people say or do things that did not happen in reality.
Long quite unconvincing, these deepfakes have become very realistic with the rise of so-called generative AI, which makes it possible to produce all kinds of content, including video, on a simple request in everyday language.
Technology from OpenAI, the creator of ChatGPT and industry leader, allows for example the Spotify platform to translate podcasts directly into another language, retaining the voice of the hosts, without additional recording or dubbing. But the use of generative AI for illegal purposes, such as fraud or disinformation, worries many authorities. The European Union hopes to finalize before the end of the year the first legislation in the world aimed at regulating these innovations. However, it will not come into force before 2026. Meta, the parent company of Instagram, did not immediately react to a request from AFP.